50th Lunar and Planetary Science Conference 2019 (LPI Contrib. No. 2132) 1859.pdf

SHARAD MAPPING OF THE CALDERA OF ARSIA MONS. I. Ganesh1, L. M. Carter1 and I. B. Smith2, 1University of Arizona, Tucson, Arizona, USA ([email protected]), 2York University, Toronto, Canada.

Introduction: Among the Tharsis Montes, only subsurface interfaces in order to avoid mapping clutter Arsia Mons shows signs of intra-caldera volcanism [1, from off-nadir relief. 2]. The caldera floor hosts several small shields, vents, Mapping results: The reflectors (indicated by col- fissures and a network of lava flows dated to be 130 ored lines in Figure 1) do not lie close to the vent field Ma old [3, 4]. Models of recurrence rates of volcanism (indicated by yellow triangles in Figure 1). The based on the surface vents constrain the intra-caldera mapped interfaces are grouped into two based on their activity between 200–300 Ma and 90–10 Ma with vent geographic location and subsurface structure. (1) The creation rates peaking at 150 Ma [5]. The possibility of reflectors in the west show layering indicating radar additional vents buried beneath the surface flows or a backscattering from multiple vertically stacked dielec- change in the style of volcanism from explosive to ef- tric interfaces. (2) The southern reflectors have high fusive eruptions ~150 Ma causes uncertainties in the backscatter values almost equal to surface reflection, increasing trend in vent production rate prior to 150 but do not show layering (Figure 2). The region be- Myr [5]. Data from SHAllow RADar (SHARAD) can tween the surface echo and the subsurface echo in the help in identifying and mapping buried volcanic depos- radargram is dark and does not show any reflections. its of distinct physical or petrochemical properties within the caldera. The dielectric structure of the sub- surface as observed by SHARAD can be useful in elu- cidating the volcanic history of the caldera.

Figure 2: SHARAD radargrams (top) and clutter simulations (bottom) from track 13952 (left) and 26413 (right). The in- terface depths have been calculated assuming a dielectric constant (relative permittivity) value ε’ = 7. Bayesian inversion of SHARAD measurements: A Bayesian approach has been used to invert equations Figure 1: THEMIS day-IR image showing the locations of of time delay and return power to estimate the loss tan- subsurface interfaces: Orange - dipping interfaces close to gent tanδ and relative permittivity ε’ of the subsurface the surface; Maroon - flat interfaces underlying those shown media. Posterior probability distribution of ε’ and tanδ in yellow; Blue - a second layer of flat interfaces directly were calculated using a maximum likelihood approach beneath the first layer. Green - bright interfaces that occur [9] without making any explicit assumptions about the close to the South caldera wall. Bright yellow triangles mark the locations of volcanic vents and fissures in the caldera. media. The resulting distribution was then sampled by SHARAD mapping: SHARAD is a sounding radar implementing a Markov chain Monte-Carlo (MCMC) onboard the Mars Reconnaissance Orbiter (MRO) that technique using the Python package emcee [10] (result- operates at a 10 MHz bandwidth centered at 20 MHz ing fit and posterior probability distribution are shown [6]. This allows for range resolution of 15 m in free in Figure 3) to identify suites of solutions that are com- space and 5–10 m in geological medium [7]. SHARAD patible with the signal return measured by SHARAD. observations are processed into radargrams that display Dielectric properties of the subsurface: the return power as a function of signal return time Western interfaces. The layered interfaces in the along the y axis and along-track distance in the x direc- west show increasing permittivity with depth. The top- tion. Radargrams from 27 night-time SHARAD tracks most layers are 30-50 meters thick and the next set of crossing the caldera were surveyed to locate late time interfaces lie 40-60 meters below the surface. The delay echoes. Only those reflectors that were not ap- permittivity values are determined to be ε’ = 4.3 ± 0.6 parent in the clutter simulations [8] were mapped as for the upper layer, ε’ = 6.6 ± 0.2 for the second layer from the top and ε’ = 8.2 ± 0.4 for the lowest discerni- 50th Lunar and Planetary Science Conference 2019 (LPI Contrib. No. 2132) 1859.pdf

ble layers (or half-space). The upper layer has a loss permittivity for the upper layer combined with a thick- tangent value of tanδ = 0.019 ± 0.003 and the lower ness > 50 meters are consistent with deposits of volcan- medium shows a higher loss tangent value of tanδ = ic ash, pumice or tuff [11]. The upper layers have loss 0.035 ± 0.005. The permittivity ε’ of a medium can be tangent values of tanδ = 0.04 ± 0.01 indicating higher linked to the medium’s bulk density ρ through the rela- power loss compared to most of the western layers. tion ε’ = 1.96ρ [11]. The increasing permittivity with Implications for volcanism in the caldera: Map- depth therefore indicates an increase in bulk density of ping of the subsurface interfaces and Bayesian model- the subsurface layers. The permittivity estimations like- ing of SHARAD signal propagation reveal differences ly represents a subsurface structure with a thicker low- in the properties of subsurface interfaces at different density medium overlying a relatively thinner high- locations within the caldera. This could be due to dif- density layer. ferences in the style of volcanism through time in these two regions resulting in different types of buried de- (A) posits. The thickness of the surface dust layer could also influence the amount of signal transmitted through [12] and consequently alter the depth to which SHARAD can detect subsurface interfaces. The higher loss tangent in the south may also preclude the detec- tion of any layers below the first interface. The differences in the reflector properties (thick- ness, dielectric permittivity) most likely represent vari- ations in eruption styles in the past. The lower layers in the western part of the caldera are possibly dense lava flows buried beneath a lower density medium that could be a vesicular lava flow or a mixture of effusive and explosive products. The unusually low permittivity values in the south are more indicative of ash or tephra overlying lava flows. The range of loss tangent values

(B) at both these locations are consistent with volcanic material. High values of dielectric loss for the lower layers in the west and the southern layers could be due to higher amounts of iron-bearing minerals in the sub- surface [13]. Alternatively, volumetric scattering in the subsurface could also cause high loss tangent values. The differences in the subsurface structure provide confirmation that the style of volcanism was not uni- form spatially and/or temporally within the caldera. References: [1] Carr, M. H. et al. (1977) JGR, 82, 3985–4015. [2] Crumpler, L. S. and Aubele, J. C. (1978) Icarus, 34, 496-511. [3] Werner, S. (2009) Ica- rus, 201, 44–68. [4] Neukum, G. (2004), Nature, 432, 971–979. [5] Richardson, J. A. et al. (2017) EPSL, 458, 170–178. [6] Seu, R. et al. (2007) JGR, 112, E05S05. [7] Carter, L. M. et al. (2009) GRL, 36, L23204. [8] Choudhary, P. et al. (2016) IEEE GRSL, Figure 3: (A) Power loss as a function of depth for a 2nd layer reflector in track 22378. The brown circles are 13, 1285–1289. [9] Fisher, R. A. (1912), Messenger SHARAD data points; the gray lines are 100 randomly cho- Math., 41, 155–160. [10] Foreman-Mackey, D. et al. sen samples from the MCMC model results plotted as linear (2012), arXiv:1202.3665 [astro-ph.IM]. [11] Ulaby, F. fits to the data points. (B) Corner plot showing the 1-D and T. et al. (1988), Microwave dielectric spectrum of 2-D posterior probability distribution of the fit parameters rocks, Rep. 23817-1-T. [12] Simon, M. N. et. al. (slope and intercept of the linear fit). (2014) JGR, 119, 2291–2299. [13] Olhoeft G. R. Southern interfaces. The bright reflectors in the (1998), 7th Intl. Conf. on GPR, 177–182. south are 50-80 meters deep and have a very low per- Acknowledgements: This work was partially fund- mittivity layer with ε’ = 2.9 ± 0.1 above and a layer ed by a SHARAD Co-I grant to L. M. Carter. with permittivity of ε’ = 7.5 ± 0.3 below. The low